Durham Research Online

Deposited in DRO: 06 April 2016 Version of attached le: Accepted Version Peer-review status of attached le: Peer-reviewed Citation for published item: Gauld, N.R. and Campbell, R.N.B. and Lucas, M.C. (2013) 'Reduced ow impacts salmonid smolt emigration in a river with low-head weirs.', Science of the total environment., 458-460 . pp. 435-443. Further information on publisher's website: http://dx.doi.org/10.1016/j.scitotenv.2013.04.063

Publisher's copyright statement: NOTICE: this is the author's version of a work that was accepted for publication in Science of The Total Environment. Changes resulting from the publishing process, such as peer review, editing, corrections, structural formatting, and other quality control mechanisms may not be reected in this document. Changes may have been made to this work since it was submitted for publication. A denitive version was subsequently published in Science of The Total Environment, 458-460, August 2013, 10.1016/j.scitotenv.2013.04.063.

Additional information:

Use policy

The full-text may be used and/or reproduced, and given to third parties in any format or medium, without prior permission or charge, for personal research or study, educational, or not-for-prot purposes provided that: • a full bibliographic reference is made to the original source • a link is made to the metadata record in DRO • the full-text is not changed in any way The full-text must not be sold in any format or medium without the formal permission of the copyright holders.

Please consult the full DRO policy for further details.

Durham University Library, Stockton Road, Durham DH1 3LY, United Kingdom Tel : +44 (0)191 334 3042 | Fax : +44 (0)191 334 2971 https://dro.dur.ac.uk 1 Reduced flow impacts salmonid smolt emigration in a river 2 with low-head weirs

3

4 Final version accepted for publication in Science of the Total Environment

5

6 N. R. Gaulda,b, R. N. B. Campbellb, M. C. Lucasa

7

8 a School of Biological and Biomedical Sciences, Durham University, South Road, Durham, UK, DH1 9 3LE

10 b The Tweed Foundation, Drygrange Steading, Melrose, UK, TD6 9DJ

11 Corresponding author: N. R. Gauld, School of Biological and Biomedical Sciences, Durham University, 12 South Road, Durham, UK, DH1 3LE

13 Email address: [email protected]

14

15 Abstract

16 The impacts of large dams on the hydrology and ecology of river systems are well 17 understood, yet the impacts of low-head structures are poorly known. While impacts of small weirs 18 on upstream-migrating fish have long been mitigated by fish ladders, it is assumed that downstream 19 migration of surface-oriented fishes is unaffected under natural flow regimes. To test this, the 20 effects of low-head weirs and the influence of river flow on the migration of brown trout (Salmo 21 trutta) smolts in the , UK, was examined. Movements of acoustic tagged smolts were 22 quantified in 2010 and 2011 using automatic listening stations and manual tracking throughout the 23 migration route. In both years smolts exhibited major losses, most likely due to predators, with 24 escapement rates of 19% in 2010 and 45% in 2011. Loss rates were greater in 2010 when flows were 25 frequently below Q95 (20% of study period), compared to 2011 when more typical flows 26 predominated (0% of study period below Q95). Smolts experienced significantly longer delay at 27 weirs during 2010 than 2011, associated with the different hydrographs during emigration as well as 28 weir design. Flow comparisons within the study periods and historical records shows that low flows 29 experienced in 2010 were not unusual. The swimming behaviour of smolts in relation to flow 30 conditions differed between years, with smolts in 2010 increasing their rate of movement in relation 31 to increasing flow at a faster rate than smolts in 2011. This is the first study to demonstrate river 32 flow impacts on the migration success of wild salmonid smolts at small weirs. Because small weirs 33 are common in rivers and because spring-summer low flow periods may become more frequent with 34 climate change (based on UKCIP09 models) and altered river hydrology, further research and 35 improved management is needed to reduce the impacts of low river flows in combination with low- 36 head weirs on salmonid smolt migration.

37 Keywords: Salmo trutta, smolt migration, habitat fragmentation, river obstructions, low flow 38 1. Introduction

39 In many developed countries there is a long history of river modification and, as a result, in-river

40 structures such as dams and weirs are present in half of the world’s rivers (Dynesius and Nilsson,

41 1994; Nilsson et al., 2005). Such modification has been integral to human population growth through

42 processes such as flood defence; power generation and farming in floodplains (Nilsson et al., 2005;

43 Poff and Hart, 2002). However, in-river barriers such as dams and weirs have a major role in the

44 fragmentation of fluvial ecosystems (Dynesius and Nilsson, 1994; Fullerton et al., 2010; Jungwirth,

45 1998; Kemp and O'Hanley, 2010). In-river barriers can have major impacts on fish populations by

46 preventing or restricting movement to habitats required for essential stages of fish life history

47 (Branco et al., 2012; Lucas and Baras, 2001; Lucas and Batley, 1996; Lucas et al., 2009; Wollebaek et

48 al., 2011). In-river barriers not only impact fish populations by restricting essential movement, there

49 is also major impacts on fish habitat due to alteration of the downstream flux of water and

50 sediment, nutrient movement, and water temperatures within rivers (Poff and Hart, 2002). The

51 effects of migration obstacles depend on factors such as fish species; river hydrology and barrier

52 type, with effects varying from short delays to complete blockage (Kemp and O'Hanley, 2010;

53 Northcote, 1998). In Europe, legislation such as the Water Framework Directive (WFD; 2000/60/EC)

54 requires free passage for migratory fish travelling between areas of river essential for their life

55 history, such as juvenile emigration from natal areas and adult spawning migrations. Failure to

56 comply can result in the river being assigned less than “Good ecological status” and may result in

57 sanctions.

58 The seaward migration of juvenile anadromous salmonids (smolts) is a crucial event in their

59 life history. Smoltification is a period of great morphological, behavioural and physiological change

60 when juvenile salmonids develop various adaptations that enable them to survive at sea (Debowski

61 et al., 1999a; Debowski et al., 1999b; Denton and Saunders, 1972; Lysfjord and Staurnes, 1998;

62 McCormick et al., 1998). The smolt migratory period is precisely timed with photoperiod, river

63 discharge and temperature playing determinate roles in its commencement (Björnsson et al., 1995; 64 Björnsson et al., 2010; McCormick, 1994; McCormick et al., 2000; McCormick et al., 2007;

65 McCormick et al., 2002). Throughout migration smolts are subject to elevated predation risk from

66 mammalian; avian and fish predators (Aarestrup et al., 1999; Aarestrup and Koed, 2003; Carss et al.,

67 1990; Dieperink et al., 2002; Dieperink et al., 2001; Harris et al., 2008; Heggenes and Borgstrom,

68 1988; Koed et al., 2002; Steinmetz et al., 2003; Svenning et al., 2005a; Svenning et al., 2005b; Wiese

69 et al., 2008). Delays at river obstructions during such a timing-specific and vulnerable life history

70 stage can potentially have large impacts on the survival of smolts and the health of salmonid stocks

71 as a whole.

72 The impacts of large dams on the hydrology and ecology of temperate river systems,

73 including downstream fish passage, especially of economically important salmonids, are relatively

74 well known. In general downstream salmonid passage efficiency over dams is high (74.6%) based on

75 recent quantitative assessment (Noonan et al., 2012). However, high smolt mortalities due to both

76 physical damage and predation have been observed at major impoundments and hydro-power

77 facilities (Aarestrup et al., 1999; Hockersmith et al., 2003; Keefer et al., 2012; Muir et al., 2001a;

78 Muir et al., 2001b; Raymond, 1979; Raymond, 1988; Smith et al., 2006; Smith et al., 2002; Williams

79 et al., 2001). Low flows due to regulation in river reaches also cause delays in smolt emigration and

80 result in increased duration of exposure to mortality risks (Aarestrup and Koed, 2003; Keefer et al.,

81 2012). However, the impacts of low-head structures, such as simple overflow weirs are poorly

82 known for downstream migrants (Lucas and Baras, 2001) with the exception of bottom-orientated

83 freshwater eels (Acou et al., 2008). While impacts of small weirs on upstream-migrating fish (Lucas

84 and Frear, 1997; Ovidio and Philippart, 2002) have been partially mitigated by fish ladders designed

85 specifically to assist upstream passage (Clay, 1995), average passage efficiencies are relatively low

86 (41.7%) (Noonan et al., 2012) and presence of passage fascilities is not always guaranteed to

87 mitigate passage concerns (Roscoe and Hitch, 2010). However, it is generally assumed that

88 downstream migration of wild surface-oriented fishes such as salmonid smolts is relatively

89 unaffected and that they will pass simple overflowing weirs unhindered under reasonably natural 90 flow regimes (Lucas and Baras, 2001). Some studies on passage of hatchery-reared smolts past small

91 weirs, in particular that of Aarestrup and Koed (2003), strongly contradict this. To test this

92 assumption for wild fish, the effects of low-head weirs and the influence of natural variations in river

93 flow on the migration behaviour and survival of anadromous brown trout (Salmo trutta) smolts were

94 examined in the River Tweed, UK, a catchment with very strong wild migratory salmonid stocks.

95 2. Study areas

96 The study was carried out on the River Tweed in southern , which drains west to east and

97 empties to the North Sea. The Tweed is the sixth largest river in mainland Britain and the second

98 largest in Scotland and has some of the largest Atlantic salmon (Salmo salar) and anadromous brown

99 trout populations in the UK (Gardiner, 1989; Sheail, 1998). The Tweed catchment covers 5000 km2

100 with an estimated 2160 kilometres of the main channel and tributaries accessible to fish (Gardiner,

101 1989). The water quality of the river is very high, with there being very little pollution present

102 (Currie, 1997). The River Tweed is a designated Site of Special Scientific Interest (SSSI) within the UK

103 and is an EU Special Area of Conservation (SAC) for Atlantic salmon and lampreys. Compared to

104 many rivers, there are relatively few anthropogenic impacts and the hydrology, although modified,

105 retains high natural variability in discharge. Several low-head engineered structures occur within the

106 River Tweed’s main channel, downstream of one of the key spawning tributaries, the ,

107 as well as in the Ettrick itself (Figure 1). The Ettrick is a regulated river and its main tributary the

108 Yarrow Water is also regulated at its outflow from St Marys Loch, 23 km upstream of its confluence

109 with the Ettrick. The average annual flow on the Yarrow is 5.58 m3 s-1, while on the Ettrick it is 15.1

110 m3 s-1 and their combined catchment areas come to 501 km2. The course of the river under

111 investigation is characterised by multiple low-head structures which are remnants of light industry,

112 most of which are now redundant (Figure 1, Table 1)

113 -Figure 1 here- 114 -Table * here-

115 3. Methods

116 3.1. Smolt capture and tagging

117 Trout smolts were captured in a trap on the Yarrow between the 1st of April and the 1st of June in

118 2010 and 2011. The smolt trap consisted of a meshed box trap placed in the outwash of the smolt

119 and debris screen of a fish farm.

120 The smolts were removed from the trap and immediately placed in a holding tub filled with

121 highly aerated river water. The fish were placed in an induction tank and anaesthetised using

122 Phenoxyethanol (0.3 ml l-1), their fork length (mm) and weight (g) were recorded before those

123 sufficiently large for tagging (over 145 mm in fork length) were placed on a V-shaped surgical table.

124 An incision (12-14 mm) was made on the ventral side of the fish anterior to the pelvic girdle. A

125 miniature coded acoustic transmitter (either Model V7-2x, 7 mm diameter, 18 mm length, 1.4 g

126 weight in air, Vemco Ltd, Nova Scotia, Canada or Model LP-7.3, 7.3 mm diameter, 18 mm length, 1.9

127 g weight in air, Thelma Biotel AS, Trondheim, Norway) was then implanted in to the peritoneal cavity

128 through the incision. Tags were chosen to have code repeat periods of 20-60 seconds and estimated

129 lives of 100 days. The incision was closed with three independent sutures (4-0 Vicryl Rapide, Ethicon

130 Ltd, Livingston, UK). The gills were aspirated with a mixture of dilute Phenoxyethanol and river water

131 during the early stages of the procedure before switching to 100% river water during the later stages

132 of the procedure. All tagging was carried out under UK Home Office License and complied with the

133 UK Animals (Scientific Procedures) Act 1986.

134 Once the procedure was complete the fish were returned to a recovery tub filled with highly

135 aerated water. When recovered the fish were placed in a keep box in the intake channel overnight

136 before release into the river; no mortalities occurred during these procedures. Details of the fish

137 released in the two seasons are given in Table 2. There was no significant difference between the 138 lengths of smolts tagged in 2010 and 2011 (Mann-Whitney U; n=103, Z=-0.445,p>0.05). Release was

139 always in groups that included untagged fish (since smolts migrate in aggregations), within 24 hours

140 of tagging, in to a section of the river 100 m below the point of capture. Due to high losses of tagged

141 smolts within the upper study section in 2010, tagged smolts were released at two additional release

142 sites, one 2 km below the point of capture and another 200 m downstream of the the Murray Cauld

143 as a way to test the impact of the weir on migration in 2011 (Table 2, Figure 1). The Murray Cauld is

144 the only intact in-river structure on the migration route and so has only a fish pass as an alternative

145 to passage over its crest. The lengths of smolts in the three release groups in 2011 were not

146 significantly different (Kruskall-Wallis; n=60, χ2= 1.0892, df = 2, p>0.05).

147 -Table 2 here-

148 3.2. Acoustic tracking

149 Acoustic tracking was carried out via a combination of fixed automatic listening stations (ALS) and

150 manual tracking at 69 KHz. Fixed ALS positions (Models VR2 & VR2W, Vemco Ltd, Nova Scotia,

151 Canada) were set approximately 11 km apart along the migration route. Sites were chosen to detect

152 fish as they approached cross-river weirs or other features of interest, with acoustic loggers located

153 in calm water to give reliable recording of tags, based upon field tests. Positioning of loggers at some

154 sites was limited by the availability of calm, deep water as well as site access. Logging stations at

155 weirs were located 50-100 m upstream of obstructions. In the estuary multiple stations were placed

156 in both the inner and outer estuary to give effective coverage. ALS stations were downloaded on a

157 weekly basis during the study period, these data allowed for the locations of each fish to be

158 estimated and help determine areas to target for manual tracking.

159 Manual tracking was carried out on foot using a Vemco VR100 (Vemco Ltd, Nova Scotia, Canada)

160 with a VH110 Directional Hydrophone attached (Vemco Ltd, Nova Scotia, Canada). Range testing was

161 conducted by placing a test tag in a known position and then measuring the distance at which the

162 test tag became undetectable on manual tracking equipment, this was repeated in several different 163 river sections with varying hydromorphological conditions. In field tracking conditions, with the

164 hydrophone kept fully submerged, the range varied between 100 m in deep pools to less than 10 m

165 in fast flowing riffles. Fish locations were recorded by the VR100 inbuilt GPS unit and later stored in a

166 GIS database.

167 In 2010, 10 tags were deployed in mesh bags in the river to estimate tag failure rate. As a further

168 control, 10 tags were deployed loose on the river bed to determine whether, and under what

169 circumstances, tags lost by fish, or following predation and subsequent tag egestion, were moved

170 passively by flows and what their detectability was.

171 3.3. Environmental data

172 River flow is recorded along the smolt migration route at the Philiphaugh gauging station of the

173 Scottish Environment Protection Agency (SEPA) on the lower Yarrow and also at their Lindean

174 (Ettrick), Boleside and Sprouston (Both Tweed) and at the Norham gauging station of the

175 Environment Agency of England and Wales (EA)(Figure 1). Historic flow records for these stations

176 were obtained from the Centre for Ecology and Hydrology (CEH) National River Flow Archive (NRFA).

177

178 4. Results

179 4.1. Inter-annual variations in survival out to sea and passage efficiencies at weirs

180 Through the combined use of stationary ALS receivers and manual tracking, survival estimates were

181 calculated for the 43 tagged smolts released in 2010 and the 60 released in 2011. The approximate

182 distance travelled by each smolt was measured from its last known location. Tags that were either

183 missing after repeated manual tracking trips or repeatedly found at the same site, without any

184 movement on successive manual tracking trips were assumed to be smolt mortalities. In total, seven

185 tags in 2010 and three tags in 2011 were assumed to be dead after repeatedly being found in the

186 same location in the river. Conversely, 28 tags in 2010 and 30 tags in 2011 were assumed to have 187 been removed from the system by terrestrial predators after a cessation in logged movements and

188 not being detected after several manual tracking trips. All of the tags deployed in the river as

189 controls in retrievable mesh bags operated for their expected durations and 90% of the tags

190 deployed loose on the river bed could be detected over their study period, none moving more than

191 1 m.

192 In 2010 only 19% of the 43 released smolts were detected leaving the river on the outer

193 estuary logger whereas 45% of the 60 released smolts reached there in 2011. One notable difference

194 between years was the variation in mortality around the Murray Cauld; in 2010 a 44% decline in

195 survival was observed there compared to a 9% decline in 2011 (Figure 2). There was a slight variation

196 in survival out to sea for release sites A and B (above the Murray Cauld) and C (below it) in 2011,

197 which had relatively normal flow, with 40%; 55% and 40% survival being observed respectively

198 (Figure 2). In 2010 there was a significant difference in smolt length between successful migrants

199 and unsuccessful migrants, with successful smolts being larger (Mann-Whitney U; n=43, Z=-2.07,

200 p=0.044). This trend may be a result of the low number of successful smolts compared to the much

201 larger number of unsuccessful smolts. However, In 2011 there was no difference in length between

202 successful and unsuccessful migrants (Mann-Whitney U; n=60, Z =-0.647, p>0.05).

203 For both years a significant negative relationship between distance travelled from release

204 site and cohort survival was recorded (2010: linear regression; n=43, R2= 0.495, F= 12.064, p= 0.005;

205 Figure 2, 2011: linear regression; n=60, R2=0.84, F=84.731, p<0.001; Figure 2). For all three release

206 sites in 2011 there were significant negative relationships between the distance travelled from

207 release sites and cohort survival (release site A: linear regression; n=20, R2=0.52, F=15.263, p=0.002;

208 Figure 2, release site B: linear regression; n=20, R2=0.72, F=37.305, p<0.001; Figure 2, release site C:

209 linear regression; n=20, R2=0.73, F=25.536, p=0.001; Figure 2). Subsequently, two of the smolts

210 tagged in 2011 were detected 20 km up the estuary of the River Tees on an acoustic array associated

211 with a separate study. The Tees estuary is approximately 144 km south of the Tweed estuary, along 212 the North Sea coast, and the tags were detected for periods of 4.3 and 60.4 hours, after respective

213 periods of 20 and 10 days following escapement from the Tweed estuary. These detections fit in

214 with prior Carlin tag data from the Tweed that shows smolts moving down the UK coastline close to

215 shore and in neighbouring estuaries (Campbell, unpublished data).

216 The passage efficiencies at three different weirs differed between years, at Murray Cauld

217 passage efficiency differed markedly between years with 46% and 100% passage efficiency being

218 observed in 2010 and 2011 respectively. Differences in passage efficiency between 2010 and 2011

219 were also observed on the other two weirs studied but were not as pronounced (Table 3). What is

220 important to note is that weir design differs between all three weirs and Murray Cauld is the only

221 fully intact weir.

222 -Figure 2 here-

223 4.2. The delay of smolts during seaward migration in 2010 and 2011 and its impact on

224 smolt movement rate

225 When comparing the mean ground speeds of migrating smolts in 2010 and 2011, using the first

226 detection of each smolt on each ALS position along the migration route and factoring in each river

227 section in to the analysis, a significant difference was observed (ANOVA; n=205, F=5.673, p<0.001;

228 Figure 3) with smolts in 2011 moving significantly faster along the migration route. Ground speed

229 data for 2011 in the river sections between release site B and logging station 1 as well as release site

230 C and logging station 2 were not included in the analysis due to the stated release sites not being

231 used in 2010.

232 Records of the migration delays experienced by smolts at localities in both 2010 and 2011

233 were retrieved from stationary ALS receivers. Delay was quantified by the duration of time between

234 the first recording and the last recording on an ALS for each tagged smolt. Data from station 5 were

235 not included, since this logger was inefficient due to noise resulting from its suboptimal location. In 236 general, smolts experienced more delay in 2010 than 2011. Smolts were more significantly delayed

237 in 2010 compared to 2011 on all freshwater ALS stations; station 1 (Mann-Whitney U; n=54, Z=-5.0,

238 p<0.001; Table 3), station 2 (Mann-Whitney U; n=47, Z=-2.33, p=0.02; Table 3), station 3 (Mann-

239 Whitney U ; n=32, Z=-2.712, p=0.011; Table 3), station 4 (Mann-Whitney U; n=19, Z=-2.966, p=0.002;

240 Table 3), station 6 (Mann-Whitney U; n=23, Z=-3.244, p=0.001; Table 3) and station 7 (Mann-

241 Whitney U; n=34, Z=-2.315, p=0.02; Table 3). However, there was no significant difference in delay in

242 the Tweed estuary between 2010 and 2011 (Mann-Whitney U; n=33, Z=-0.336, p>0.05; Table 3),

243 suggesting that either the factors influencing delay within the river were not present or were of less

244 importance within the estuary or that a different set of factors govern estuarine movements.

245 -Table 3 here-

246 -Figure 3 here-

247 4.3. Variation in flow conditions between 2010 and 2011 and its influence on smolt ground

248 speed

249 Using mean daily flow data retrieved from SEPA and the EA and flow duration curves from the CEH

250 NRFA, the flow conditions along the migration route during the typical smolt migration period (1

251 April to 30 June) in 2010 and 2011 were analysed. The Lindean SEPA gauging station was used as a

252 proxy for the flow at the Murray Cauld as it is approximately 6 km downstream from the weir and

253 there are no large tributaries joining the Ettrick in this section of river. The two years’ flows at

254 Lindean, during the key migration period, differed markedly, with mean daily flows declining below

255 the Q95 flow for 18 days in 2010 and not at all in 2011. There were several high flow events in 2011

256 whereas the only flow increases in 2010 were the results of artificial weekly freshets from St Mary’s

257 Loch on the Yarrow system (Figure 4).

258 -Figure 4 here- 259 Using historical flow records from the CEH NRFA for Lindean extending back to 1962 the

260 prevalence of daily flows under Q95 was calculated for each year in the 49 year period. Days where

261 flow was low there during the migration period were not uncommon (Figure 5). Short periods of

262 flow restriction occurred frequently and periods where at least 15 days out of the 90 day period

263 were below Q95 daily flows occurred at least once a decade (Figure 5). There have therefore been

264 periods of flow restriction similar to that experienced in 2010 previously and they are likely to

265 reoccur.

266 -Figure 5 here-

267 The influence of flow conditions on smolt migration speed was calculated from the net

268 ground speed of individual smolts between two successive ALS positions using the first record of

269 each smolt at each ALS as it moved downstream and then matching the speed to the mean flow

270 conditions during the period of transit using 15-minute gauged flows from the nearest SEPA flow

271 gauging stations to the fixed ALS positions. This was carried out for all sequential pairs of ALSs. For

272 both years a positive relationship between elevated flow (m3s-1) and increased net ground speed (km

273 h-1) was observed; 2010 (Regression; n=88, R=0.719, p<0.001; Figure 6), 2011 (Regression; n=218,

274 R=0.579, p<0.001; Figure 6). However, when the relationships between net groundspeed and mean

275 flow were compared between years using an ANCOVA there was a highly significant difference in

276 slope (n=306, F=147.73, p<0.001). These results suggest that smolts released in 2010 undertook

277 increasingly more active swimming within the flows in which they exhibited downstream migration

278 than the smolts released in 2011.

279 -Figure 6 here-

280

281

282 283 5. Discussion

284 This study shows, for the first time, that surface-orientated wild fishes, migrating

285 downstream, can be markedly impeded by small overflowing weirs, and that the effects of this are

286 dramatically increased during low-flow conditions. These delays are associated with losses of

287 migrating fishes, again substantially elevated during low-flow conditions. While these effects are

288 known for salmonids at large impoundments, especially hydroelectric dams, with or without surface

289 bypasses (Hockersmith et al., 2003; Muir et al., 2001a; Muir et al., 2001b; Raymond, 1979; Raymond,

290 1988; Smith et al., 2006; Williams et al., 2001), and also for benthically orientated eels (Acou et al.,

291 2008; Boubée and Williams, 2006; Gosset et al., 2005), they have not been recorded for wild juvenile

292 salmonids in relatively natural river systems. However, manipulative studies with smolts have shown

293 that modified surface bypasses reduce the delay in passing weirs compared to conventional

294 bypasses (Haro et al., 1998). These results strongly suggest that small obstructions can have much

295 larger than expected impacts on seaward escapement of anadromous brown trout smolts and given

296 the observation that low flows dramatically exacerbate these problems, any climate scenario (such

297 as UKCIP02 and UKCP09) that results in increased frequency of low river flows during spring and

298 early summer is a very real concern (Arnell, 2004; Christierson et al., 2012; Marsh, 2004; Wilby and

299 Harris, 2006). However, it is possible that climate change may bring an increase in water availability

300 for the UK in some scenarios (IPCC SRES A2 and B2) (Xenopoulos et al., 2005).

301 The results from the automated acoustic tracking of the smolts migrating to the sea in 2010

302 and 2011 clearly showed a disparity in the degree to which they were delayed in different river

303 sections between the two seasons. These also showed that obstructions in river sections, such as

304 weirs, also exacerbate delays during periods of reduced river flow. In general very little work has

305 been conducted to link overflowing barriers to the passage and behaviour of freshwater fish during

306 downstream movement. In Australian studies Murray cod (Maccullochella peelii) and golden perch

307 (Macquaria ambigua) displaced above weirs displayed a reluctance to move past low-head weirs

308 when attempting to home downstream (O'Connor et al., 2006). Negative impacts of weirs were also 309 observed in hatchery reared Atlantic salmon and anadromous brown trout smolts released in small

310 Danish rivers where they suffered from increased delay and mortality in proximity to small fish farm

311 weirs (Aarestrup and Koed, 2003). Low flows spread across the breadth of obstructions such as

312 overflowing weirs spanning whole channels, give depths over their crests that are very shallow,

313 which may reduce the behavioural stimuli (one or more combinations of velocity, depth, velocity

314 gradient, turbulence) needed to get fish to continue past the barrier. Haro et al., (1998) found

315 American shad (Alosa sapidissima) to be unwilling to approach the small surface water bypasses that

316 would allow them to move downstream at large barriers, while Enders et al. (2009) demonstrated a

317 similar unwillingness for salmonid smolts under experimental conditions, showing that hydraulic

318 changes at surface bypasses do not necessarily promote effective downstream passage of surface-

319 orientated fishes.

320 In the current study it was inferred that acoustic tag loss was very likely due to removal of

321 tagged fish from the river by terrestrial predators because; 1) transmitters were lost well within the

322 quoted lifetime of the tags; 2) control transmitters deployed in the river showed zero failure rate

323 within the quoted life; 3) loose control tags on the river bed could be reliably detected by tracking

324 gear and moved little and, 4) predation by aquatic predators (in this study area, large brown trout),

325 would have resulted in acoustic tags being retained in the aquatic environment and detectable. The

326 most common avian predators on the Tweed are goosander (Mergus merganser) and grey heron

327 (Ardea cinerea), the former occurs in large numbers during the smolt migration season when they

328 can form large feeding aggregations. Their diet on the Tweed has been investigated by Marquiss, et

329 al (1998), who estimated their consumption of smolt-sized salmonids could be up to 4.79 per

330 goosander per day in March and April and up to 1.8 per day in May. The survival of smolts during

331 migration was radically different between the two seasons studied, that of 2010 (19%) being below

332 half that of 2011 (45%). These levels can be compared with those of conventionally tagged

333 anadromous brown trout smolts in Norway which were estimated to have a survival rate of 24% for

334 their first seaward migration (Berg and Berg, 1987) and with the survival of chinook salmon 335 (Oncorhynchus tshawytscha) smolts migrating down the Snake and Columbia rivers where survival to

336 the sea was estimated to be around 27.5% (Welch et al., 2008). However, the Columbia River system

337 is of much greater size and has much larger impoundments than the Tweed catchment.

338 The mortality of Atlantic salmon smolts during in-river migration has been estimated for

339 several different rivers in previous studies. Overall mortality, calculated on a kilometre by kilometre

340 basis ranged from 0.3 to 5% per kilometre (Davidsen et al., 2009; Dieperink et al., 2002; Koed et al.,

341 2002; Martin et al., 2009; Moore et al., 1998; Thorstad et al., 2012a; Thorstad et al., 2012b). In

342 comparison anadromous brown trout smolts tracked in the Tweed in 2010 and 2011 suffered 0.88%

343 and 0.55% mortality per km respectively, well within the range of mortality observed for salmon. It is

344 important to note that these studies only included the lower reaches and estuary of their rivers

345 where predation is expected to be more intense while the present study examined migration over

346 100.29 km of river and estuary.

347 Mortality at individual weirs during migration varied within and between years, with

348 mortality ranging between 2-44% per cohort of fish arriving at each weir with an ALS near it (the

349 Murray Cauld, Melrose Cauld and Mertoun Cauld) in 2010 and 5-9% in 2011. In comparison, stocked

350 brown trout smolt mortality at various fish farm weirs in Denmark varied between 15-64%, although

351 it is important to note that piscivorous predators such pike (Esox lucius) and zander (Sander

352 lucioperca) are present in Danish rivers (Aarestrup and Koed, 2003) but are absent in the studied

353 section of the River Tweed. Passage efficiencies at these weirs also varied between 46-90% in 2010

354 and 92-100% in 2011. Murrays Cauld was particularly inefficient in 2010 with downstream passage

355 efficiency being only 46%, well below the average downstream passage efficiency of 68.5% seen in

356 Noonan et al., (2012). This low efficiency during low flow periods is most probably the consequence

357 of Murray Cauld being the only fully intact weir along the migration route, with other weirs either

358 being in a ruinous state or cut. 359 The flow conditions in the period of study were markedly different between years. The April

360 to June water levels of 2010 were characterised by low flows that dipped below Q95 for a total of 18

361 days whilst the 2011 flows for the same period exceeded Q10 flows for two consecutive days during

362 the largest spate and had other elevated periods. From a historical perspective, low flows similar to

363 those that were prevalent in 2010 for the study period have been recorded regularly on the Ettrick

364 between 1962 and 2011. The use of Q95 flows as an estimation of low flows is now widely practised

365 in Europe (Gustard et al., 1992; Laaha and Blöschl, 2007; Smakhtin, 2001). Studies into the migration

366 of chinook salmon on rivers with large barriers have shown a positive relationship between

367 increased river flow and increased smolt survival during migration (Connor et al., 2003; Smith et al.,

368 2003).While the Tweed is a much smaller river, with small barriers, the same pattern is apparent –

369 higher smolt mortality in seasons with low flows and vice-versa.

370 Smolt swimming speed increased in relation to flow in both years of the study. However,

371 smolts in 2010 showed a steeper relationship of ground speed to river discharge than smolts in

372 2011. This may be a consequence of the overall lower flow conditions in the river in 2010 compared

373 to 2011 possibly meaning that smolts moving downstream in 2010 did so more actively than smolts

374 released in 2011. Conversely, smolts in 2011 displayed more active swimming behaviour at lower

375 flow levels than smolts in 2010, this is possibly due to smolts in 2011 not suffering the same flow

376 restriction as smolts in 2010 and therefore movement may not be as impeded by in river structures.

377 Similarly, previous research into anadromous brown trout and Atlantic salmon smolt migration has

378 also found a correlation between river discharge and smolt net ground speeds (Aarestrup et al.,

379 2002; Martin et al., 2009). Smolt ground speeds were low in sections from release to detections

380 upstream of Philiphaugh weir in both 2010 and 2011, but these low speeds include periods during

381 which smolts may have been preparing to emigrate and exhibited holding behaviour.

382 The conclusion of this study is that passage of downstream-migrating salmonid smolts is not

383 only impacted by the large dams with which river managers are familiar, but probably also by much 384 smaller low head weirs that Lucas et al. (2009) report as being much more abundant and which

385 impound water and create zones of reduced flow rate. Current passage provision for downstream-

386 migrating salmonid smolts is probably inadequate at many weirs and periodic low flows during the

387 smolt migratory period should be a management concern, especially for areas where salmonid

388 stocks are a highly prized economic asset. Most fish passage facilities, such as technical fish ladders,

389 are designed for upstream migrants, and while downstream fish bypasses exist, they have been little

390 used on low-head overflowing weirs and have rarely been evaluated for their efficiency (Haro et al.,

391 1998; Scruton et al., 2002, 2007). In the face of climate change and un- certain variability in river

392 flows,where low-head structures are no lon- ger needed, removal should be strongly considered

393 along with the construction of bypasses for reducing emigration delays and mortality in salmonid

394 smolts (Arnell, 2004; Christierson et al., 2012; Garcia de Leaniz, 2008; Kemp and O'Hanley, 2010;

395 Marsh, 2004; Wilby and Harris, 2006; Xenopoulos et al., 2005). To ultimately test the impact of

396 weirs, future studies should consider a tenable before–after control impact (BACI) design, using

397 multiple years worth of smolt migration data for each treatment. Further to this, more detailed

398 information on smolts lost while migrating downstream would also be very useful for management

399 purposes, unless definite causes can be assigned for losses it is difficult to take measures against

400 them.

401 Acknowledgments

402 This work was supported by the Living North Sea project and the Interreg IVB North Sea

403 Region Programme. The authors would like to thank the landowners, river boatmen and angling

404 clubs involved in the work for allowing us to place logging equipment in their waters as well as their

405 help and advice. The authors would also like to thank River Tweed Commission Bailiffs Alan

406 Davidson, Eric Hastings and Kenny Graham for help retrieving logging equipment from the estuary.

407 The authors would also like to further thank Nick Yonge, Fay Hieatt, James Hunt and Kenny Galt at

408 the Tweed foundation for their help and support during the period of work. 409 References 410

411 Aarestrup K, Jepsen N, Rasmussen G. Movements of two strains of radio tagged Altlantic salmon, 412 Salmo salar L., smolts through a reservoir. Fish. Manage. Ecol. 1999; 6: 97-107. 413 Aarestrup K, Koed A. Survival of migrating sea trout (Salmo trutta) and Atlantic salmon (Salmo salar) 414 smolts negotiating weirs in small Danish rivers. Ecol. Freshwat. Fish 2003; 12: 169-176. 415 Aarestrup K, Nielsen C, Koed A. Net ground speed of downstream migrating radio-tagged Atlantic 416 salmon (Salmo salar L.) and brown trout (Salmo trutta L.) smolts in relation to environmental 417 factors. Hydrobiologia 2002; 483: 95-102. 418 Acou A, Laffaille P, Legault A, Feunteun E. Migration pattern of silver eel (Anguilla anguilla, L.) in an 419 obstructed river system. Ecol. Freshwat. Fish 2008; 17: 432-442. 420 Arnell NW. Climate-change impacts on river flows in Britain: the UKCIPO2 scenarios. Water and 421 Environment Journal 2004; 18: 112-117. 422 Berg OK, Berg M. Migrations of sea trout, Salmo trutta L., from the Vardnes river in northern 423 Norway. J. Fish Biol. 1987; 31: 113-121. 424 Björnsson B, Stefansson S, Hansen T. Photoperiod regulation of plasma growth hormone levels 425 during parr-smolt transformation of Atlantic salmon: implications for hypoosmoregulatory 426 ability and growth. Gen. Comp. Endocrinol. 1995; 100: 73-82. 427 Björnsson BT, Stefansson SO, McCormick SD. Environmental endocrinology of salmon smoltification. 428 Gen. Comp. Endocrinol. 2010; 170: 290-298. 429 Boubée JAT, Williams EK. Downstream passage of silver eels at a small hydroelectric facility. Fish. 430 Manage. Ecol. 2006; 13: 165-176. 431 Branco P, Segurado P, Santos JM, Pinheiro P, Ferreira MT. Does longitudinal connectivity loss affect 432 the distribution of freshwater fish? Ecological Engineering 2012; 48: 70-78. 433 Carss DN, Kruuk H, Conroy JWH. Predation on adult Atlantic salmon, Salmo salar L., by otters, Lutra 434 lutra (L.), within the River Dee system, Aberdeenshire, Scotland. J. Fish Biol. 1990; 37: 935- 435 944. 436 Christierson BV, Vidal J-P, Wade SD. Using UKCP09 probabilistic climate information for UK water 437 resource planning. Journal of Hydrology 2012; 424–425: 48-67. 438 Clay CH. Design of fishways and other fish facilities. Boca Raton: Lewis Publishers, 1995. 439 Connor WP, Burge HL, Yearsley JR, Bjornn TC. Influence of Flow and Temperature on Survival of Wild 440 Subyearling Fall Chinook Salmon in the Snake River. North American Journal of Fisheries 441 Management 2003; 23: 362-375. 442 Currie J. Pollution prevention on the River Tweed: past, present and future. Sci. Total Environ. 1997; 443 194-195: 147-154. 444 Davidsen JG, Rikardsen AH, Halttunen E, Thorstad EB, ØKland F, Letcher BH, et al. Migratory 445 behaviour and survival rates of wild northern Atlantic salmon Salmo salar post-smolts: 446 effects of environmental factors. J. Fish Biol. 2009; 75: 1700-1718. 447 Debowski P, Dobosz S, Robak S. Estimation of smoltification of hatchery-reared sea trout (Salmo 448 trutta morpha trutta L.) based on body morphology. Arch. Pol. Fish. 1999a; 7: 257-266. 449 Debowski P, Glogowski J, Robak S, Dobosz S. Smoltification of hatchery-reared Atlantic salmon 450 (Salmo salar L.)-indices and methods of estimation. Arch. Pol. Fish. 1999b; 7: 267-279. 451 Denton EJ, Saunders RL. On the organization of silvery layers in the skin of the Atlantic salmon 452 (Salmo salar) during smoltification and on the regeneration of these layers under abnormal 453 lighting conditions. Journal of the Marine Biological Association of the UK 1972; 52: 889-898. 454 Dieperink C, Bak BD, Pedersen LF, Pedersen MI, Pedersen S. Predation on Atlantic salmon and sea 455 trout during their first days as postsmolts. J. Fish Biol. 2002; 61: 848-852. 456 Dieperink C, Pedersen S, Pedersen MI. Estuarine predation on radiotagged wild and domesticated 457 sea trout (Salmo trutta L.) smolts. Ecol. Freshwat. Fish 2001; 10: 177-183. 458 Dynesius M, Nilsson C. Fragmentation and flow regulation of river systems in the northern third of 459 the world. Science 1994; 266: 753-762. 460 Enders EC, Gessel MH, Williams JG. Development of successful fish passage structures for 461 downstream migrants requires knowledge of their behavioural response to accelerating 462 flow. Can. J. Fish. Aquat. Sci. 2009; 66: 2109-2117. 463 Fullerton AH, Burnett KM, Steel EA, Flitcroft RL, Pess GR, Feist BE, et al. Hydrological connectivity for 464 riverine fish: measurement challenges and research opportunities. Freshwater Biology 2010: 465 no-no. 466 Garcia de Leaniz C. Weir removal in salmonid streams: implications, challenges and practicalities. 467 Hydrobiologia 2008; 609: 83-96. 468 Gardiner R. Tweed Juvenile Salmon and Trout Stocks. In: Mills D, editor. Tweed towards 2000. The 469 Tweed Foundation, Melrose, 1989, pp. 105-114. 470 Gosset C, Travade F, Durif C, Rives J, Elie P. Tests of two types of bypass for downstream migration of 471 eels at a small hydroelectric power plant. River Res. Appl. 2005; 21: 1095-1105. 472 Gustard A, Bullock A, Dixon JM. Low flow estimation in the United Kingdom. Report No. 108. 473 Institute of Hydrology, Wallingford, 1992. 474 Haro A, Odeh M, Noreika J, Castro-Santos T. Effect of water acceleration on downstream migratory 475 behavior and passage of Atlantic salmon smolts and juvenile American shad at surface 476 bypasses. Trans. Am. Fish. Soc. 1998; 127: 118-127. 477 Harris CM, Calladine JR, Wernham CV, Park KJ. Impacts of piscivorous birds on salmonid populations 478 and game fisheries in Scotland: a review. Wildl. Biol. 2008; 14: 395-411. 479 Heggenes J, Borgstrom R. Effect of mink, Mustela vison Schreber, predation on cohorts of juvenile 480 Atlantic salmon, Salmo safar L., and brown trout, S. trutta L., in three small streams. J. Fish 481 Biol. 1988: 885-894. 482 Hockersmith EE, Muir WD, Smith SG, Sandford BP, Perry RW, Adams NS, et al. Comparison of 483 migration rate and survival between radio-tagged and PIT-tagged migrant yearling chinook 484 salmon in the snake and Columbia rivers. N. Am. J. Fish. Manage. 2003; 23: 404-413. 485 Jungwirth M. River continuum and fish migration- going beyond the longitudinal river corridor in 486 understanding ecological integrity. In: Jungwirth M, Schmutz S, Weiss S, editors. Fish 487 Migration and Fish Bypasses. Fishing News Books, Oxford, 1998, pp. 19-32. 488 Keefer ML, Taylor GA, Garletts DF, Helms CK, Gauthier GA, Pierce TM, et al. Reservoir entrapment 489 and dam passage mortality of juvenile Chinook salmon in the Middle Fork Willamette River. 490 Ecol. Freshwat. Fish 2012; 21: 222-234. 491 Kemp PS, O'Hanley JR. Procedures for evaluating and prioritising the removal of fish passage 492 barriers: a synthesis. Fish. Manage. Ecol. 2010: 297-322. 493 Koed A, Jepsen N, Aarestrup K, Nielsen C. Initial mortality of radio-tagged Atlantic salmon (Salmo 494 salar L .) smolts following release downstream of a hydropower station. Hydrobiologia 2002; 495 483: 31-37. 496 Laaha G, Blöschl G. A national low flow estimation procedure for Austria. Hydrological Sciences 497 Journal 2007: 37-41. 498 Lucas MC, Baras E. Migration of Freshwater Fishes. Oxford; Malden, MA: Blackwell Science Oxford, 499 2001. 500 Lucas MC, Batley E. Seasonal movements and behaviour of adult barbel Barbus barbus, a riverine 501 cyprinid fish: implications for river management. J. Appl. Ecol. 1996: 1345-1358. 502 Lucas MC, Bubb DH, Jang M-H, Ha K, Masters JEG. Availability of and access to critical habitats in 503 regulated rivers: effects of low-head barriers on threatened lampreys. Freshwater Biology 504 2009; 54: 621-634. 505 Lucas MC, Frear PA. Effects of a flow-gauging weir on the migratory behaviour of adult barbel, a 506 riverine cyprinid. J. Fish Biol. 1997; 50: 382-396. 507 Lysfjord G, Staurnes M. Gill Na+ –K+ -ATPase activity and hypoosmoregulatory ability of seaward 508 migrating smolts of anadromous Atlantic salmon (Salmo salar) , sea trout (Salmo trutta) and 509 Arctic char (Salvelinus alpinus) in the Hals river, northern Norway. Aquaculture 1998; 168: 510 279-288. 511 Marsh TJ. The UK drought of 2003: A hydrological review. Weather 2004; 59: 224-230. 512 Marquiss M, Carss DN, Armstrong JD, Gardiner R. Fish-eating birds and salmonids in Scotland. The 513 Scottish Office Agriculture and Fisheries Department, Edinburgh, 1998. 514 Martin F, Hedger RD, Dodson JJ, Fernandes L, Hatin D, Caron F, et al. Behavioural transition during 515 the estuarine migration of wild Atlantic salmon (Salmo salar L.) smolt. Ecol. Freshwat. Fish 516 2009; 18: 406-417. 517 McCormick SD. Ontogeny and evolution of salinity tolerance in anadromous salmonids: hormones 518 and heterochrony. Estuaries 1994; 17: 26-33. 519 McCormick SD, Hansen LP, Quinn TP, Saunders RL. Movement, migration, and smolting of Atlantic 520 salmon (Salmo salar). Can. J. Fish. Aquat. Sci. 1998; 55: 77-92. 521 McCormick SD, Moriyama S, Björnsson BT. Low temperature limits photoperiod control of smolting 522 in atlantic salmon through endocrine mechanisms. American Journal of Physiology. 523 Regulatory, Integrative and Comparative Physiology 2000; 278: R1352-61. 524 McCormick SD, Shrimpton JM, Moriyama S, Bjornsson BT. Differential hormonal responses of 525 Atlantic salmon parr and smolt to increased daylength: a possible developmental basis for 526 smolting. Aquaculture 2007; 273: 337-344. 527 McCormick SD, Shrimpton JM, Moriyama S, Björnsson BT. Effects of an advanced temperature cycle 528 on smolt development and endocrinology indicate that temperature is not a zeitgeber for 529 smolting in Atlantic salmon. J. Exp. Biol. 2002; 205: 3553-3560. 530 Moore A, Ives S, Mead TA, Talks L. The migratory behaviour of wild Atlantic salmon (Salmo salar L.) 531 smolts in the River Test and Southampton Water, southern England. Hydrobiologia 1998; 532 371/372: 295-304. 533 Muir WD, Smith SG, Williams JG, Hockersmith EE, Skalski JR. Survival Estimates for Migrant Yearling 534 Chinook Salmon and Steelhead Tagged with Passive Integrated Transponders in the Lower 535 Snake and Lower Columbia Rivers, 1993–1998. North American Journal of Fisheries 536 Management 2001a; 21: 269-282. 537 Muir WD, Smith SG, Williams JG, Sandford BP. Survival of Juvenile Salmonids Passing through Bypass 538 Systems, Turbines, and Spillways with and without Flow Deflectors at Snake River Dams. 539 North American Journal of Fisheries Management 2001b; 21: 135-146. 540 Nilsson C, Reidy Ca, Dynesius M, Revenga C. Fragmentation and flow regulation of the world's large 541 river systems. Science 2005; 308: 405-8. 542 Noonan MJ, Grant JWA, Jackson CD. A quantitative assessment of fish passage efficiency. Fish Fish. 543 2012; 13: 450-464. 544 Northcote TG. Migratory behaviour of fish and its significance to movement through riverine fish 545 passage facilities. In: Jungwirth M, Schmutz S, Weiss S, editors. Fish Migration and Fish 546 Bypasses. John Wiley & Sons, Oxford, 1998, pp. 3-18. 547 O'Connor JP, O'Mahony DJ, O'Mahony JM, Glenane TJ. Some impacts of low and medium head weirs 548 on downstream fish movement in the Murray-Darling Basin in southeastern Australia. Ecol. 549 Freshwat. Fish 2006; 15: 419-427. 550 Ovidio M, Philippart J-C. The impact of small physical obstacles on upstream movements of six 551 species of fish. Hydrobiologia 2002; 483: 55-69. 552 Poff NL, Hart DD. How dams vary and why it matters for the emerging science of dam removal. 553 Bioscience 2002; 52: 659-668. 554 Raymond HL. Effects of dams and impoundments on migrations of juvenile chinook salmon and 555 steelhead from the Snake River, 1966 to 1975. Trans. Am. Fish. Soc. 1979; 108: 505-529. 556 Raymond HL. Effects of hydroelectric development and fisheries enhancement on spring and 557 summer chinook salmon and steelhead in the Columbia River basin. N. Am. J. Fish. Manage. 558 1988; 8: 1-24. 559 Roscoe DW, Hinch SG. Effectiveness monitoring of fish passage facilities: historical trends, 560 geographic patterns and future directions. Fish Fish. 2010; 11: 12-33. 561 Scruton DA, McKinley RS, Kouwen N, Eddy W, Booth RK. Use of telemetry and hydraulic modeling to 562 evaluate and improve fish guidance efficiency at a louver and bypass system for 563 downstream-migrating Atlantic salmon (Salmo salar) smolts and kelts. Hydrobiologia 2002; 564 483: 83-94. 565 Scruton DA, Pennell CJ, Bourgeois CE, Goosney RF, Porter TR, Clarke KD. Assessment of a retrofitted 566 downstream fish bypass system for wild Atlantic salmon (Salmo salar) smolts and kelts at a 567 hydroelectric facility on the Exploits River, Newfoundland, Canada. Hydrobiologia 2007; 582: 568 155-169. 569 Sheail J. The Tweed fisheries: An historical perspective. Sci. Total Environ. 1998; 210: 469-482. 570 Smakhtin VU. Low flow hydrology: a review. Journal of Hydrology 2001; 240: 147-186. 571 Smith SG, Muir WD, Hockersmith EE, Zabel RW, Graves RJ, Ross CV, et al. Influence of river 572 conditions on survival and travel time of Snake River subyearling fall Chinook salmon. N. Am. 573 J. Fish. Manage. 2003; 23: 939-961. 574 Smith SG, Muir WD, Marsh DM, Williams JG, Skalski JR. Survival estimates for the passage of spring- 575 migrating juvenile salmonids through Snake and Columbia River dams and reservoirs, 2005. 576 Report of the National Marine Fisheries Service to the Bonneville Power Administration, 577 Portland, Oregon, 2006. 578 Smith SG, Muir WD, Williams JG, Skalski JR. Factors associated with travel time and survival of 579 migrant yearling chinook salmon and steelhead in the lower Snake River. N. Am. J. Fish. 580 Manage. 2002; 22: 385-405. 581 Steinmetz J, Kohler SL, Soluk DA. Birds are overlooked top predators in aquatic food webs. Ecology 582 2003; 84: 1324-1328. 583 Svenning M, Borgstrom R, Dehli T, Moen G, Barrett R, Pedersen T, et al. The impact of marine fish 584 predation on Atlantic salmon smolts (Salmo salar) in the Tana estuary, North Norway, in the 585 presence of an alternative prey, lesser sandeel (Ammodytes marinus). Fisheries Research 586 2005a; 76: 466-474. 587 Svenning M, Fagermo S, Barrett R, Borgstrom R, Vader W, Pedersen T, et al. Goosander predation 588 and its potential impact on Atlantic salmon smolts in the River Tana estuary, northern 589 Norway. J. Fish Biol. 2005b; 66: 924-937. 590 Thorstad EB, Uglem I, Finstad B, Chittenden CM, Nilsen R, Okland F, et al. Stocking location and 591 predation by marine fishes affect survival of hatchery-reared Atlantic salmon smolts. Fish. 592 Manage. Ecol. 2012a; 19: 400-409. 593 Thorstad EB, Whoriskey F, Uglem I, Moore A, Rikardsen AH, Finstad B. A critical life stage of the 594 Atlantic salmon Salmo salar: behaviour and survival during the smolt and initial post-smolt 595 migration. J. Fish Biol. 2012b; 81: 500-542. 596 Welch DW, Rechisky EL, Melnychuk MC, Porter AD, Walters CJ, Clements S, et al. Survival of 597 migrating salmon smolts in large rivers with and without dams. PLoS Biol. 2008; 6: e265. 598 Wiese FK, Parrish JK, Thompson CW, Maranto C. Ecosystem-based management of predator-prey 599 relationships: piscivorous birds and salmonids. Ecol. Appl. 2008; 18: 681-700. 600 Wilby RL, Harris I. A framework for assessing uncertainties in climate change impacts: Low-flow 601 scenarios for the River Thames, UK. Water Resour. Res. 2006; 42: W02419. 602 Williams JG, Smith SG, Muir WD. Survival estimates for downstream migrant yearling juvenile 603 salmonids through the Snake and Columbia rivers hydropower system, 1966-1980 and 1993- 604 1999. N. Am. J. Fish. Manage. 2001; 21: 310-317. 605 Wollebaek J, Heggenes J, Roed KH. Population connectivity: dam migration mitigations and 606 contemporary site fidelity in arctic char. BMC Evol. Biol. 2011; 11. 607 Xenopoulos MA, Lodge DM, Alcamo J, Märker M, Schulze K, Van Vuuren DP. Scenarios of freshwater 608 fish extinctions from climate change and water withdrawal. Global Change Biol. 2005; 11: 609 1557-1564. 610 611 Tables

612 Table 1: Descriptions of in river structures along the studied smolt migratory route. * 613 Structure crosses river at an angle to the flow.

Year Structure Fish pass Location (latitude, Name of Structure structure Structure head-loss present longitude ,°) structure status built width (m) (m)

Murray Cauld Intact 1847 65 3 Pool and spill 55.537667, -2.874796

Not 55.602007, -2.726349 Melrose Cauld Ruinous known 102 1 None

Mertoun Rebuilt in 55.582512,-2.623382 Cauld Cut 1990s 98 3 Pool and spill

Rutherford Not 55.57769, -2.550825 Cauld Ruinous known 153 1 None

Middle Multiple pool 55.599875,-2.439349 Kelso Cauld Cut ages 300* 2 and spill

Hendersyde Not 55.624852, -2.382158 Cauld Cut known 230 2 Pool and spill

The Lees Not 55.642852, -2.250394 Cauld Cut known 100 ca. 1 None

Coldstream 55.654607, -2.241373 bridge apron Cut 1784 96 ca. 1 None

Milne Graden Not 55.691506, -2.195022 Cauld Ruined known 98 ca. 1 None

614

615

616

617

618

619

620

621

622

623 624 Table 2: Summary data for smolts tagged in 2010 and 2011. The release sites are shown on 625 Figure 1. * Tag to body weight ratio is calculated from masses in air.

Tagging Number Fork length [mean  SD Weight [mean  SD Tag/body weight ratio Release site date tagged (range), mm] (range), g] [mean (range), %]*

Release site A 29/04/2010 14 163.2  16.5 (145-190) 45.6  15.2 (30-77) 4.5 (2.5 – 6.3)

Release site A 07/05/2010 20 161.5  15.5 (140-202) 41.4  13.4 (23-82) 5.0 (2.3 -8.3)

Release site A 13/05/2010 9 175.8  18.3 (156-200) 54.6  18.6 (29-81) 3.9 (2.3 – 6.6)

2010 Total 43 165  17 (140-202) 45.5  15.7 (23-82) 4.6 (2.3 – 8.3)

Release site A 21/04/2011 3 155  8.7 (150-165) 38  9.5 (32-49) 5.2 (3.9 – 5.9)

Release site A 22/04/2011 6 164.3  19.5 (142-199) 45.7  16.7 (31-77) 4.5 (2.5 – 6.1)

Release site A 26/04/2011 4 182.2  17 (159-198) 59.3  17.5 (35-76) 3.5 (2.5 – 5.4)

Release site A 04/05/2011 7 165  33.9 (140-220) 50.4  32.6 (23-97) 5.1 (2.0 – 8.3)

Release site A Total 20 166.7  24.3 (140-220) 48.9  22.6 (23-97) 4.6 (2.0 – 8.3)

Release site B 21/04/2011 3 160  15 (145-175) 44  11.5 (31-53) 4.6 (3.6 – 6.1)

Release site B 22/04/2011 6 161.5  20.3 (147-197) 41.8  12.5 (32-62) 4.8 (3.1 – 5.9)

Release site B 26/04/2011 4 161.5  7.3 (154-171) 42  7 (33-49) 4.6 (3.9 – 5.8)

Release site B 04/05/2011 7 170.3  16.9 (154-202) 50.3  17.7 (34-86) 4.1 (2.2 – 5.6)

Release site B Total 20 164.4  15.9 (145-202) 45.2  13.3 (31-86) 4.5 (2.2 -6.1)

Release site C 21/04/2011 3 163.3  20.2 (140-175) 43.3  13.9 (28-55) 4.8 (3.5 -6.8)

Release site C 22/04/2011 6 171.7  8.1 (160-182) 50.5  8.3 (40-62) 3.8 (3.1 – 4.8)

Release site C 26/04/2011 4 173.8  21.6 (142-190) 58.5  19.7 (31-78) 3.7 (2.4 – 6.1)

Release site C 04/05/2011 7 167.4  20.7 (145-205) 46.9  20.5 (20-85) 4.8 (2.2 – 9.5)

Release site C Total 20 169.4  16.8 (142-205) 49.8  16.1 (28-85) 4.3 (2.2 – 9.5)

2011 Total 60 166.8  19.2 (140-220) 47.9  17.6 (23-97) 4.5 (2.0 – 9.5)

626

627

628

629

630

631 632 Table 3. Delay and barrier passage efficiencies at ALS positions along the smolt migration route 633 through the river and estuary. Station 5 not listed due to insufficient sample size recorded there.

ALS Immediately In-river 2010 Delay 2011 Delay 2010 2011 Station Upstream of structure (median(Q1- Q3), (median(Q1- Q3), Passage Passage in-river characteristics minutes) minutes) efficiency efficiency structure (%) (%) 1 Yes Intact 4497.3 (109.9-25029.4) 5.8 (2.7-26.4) 46 100 2 Yes Ruinous 7.1 (1.8-18.8) 2.1 (0.9-4.6) 76 92 3 Yes Cut 1.11 (0.2- 2.7) 0.1 (0.1-0.5) 90 94 4 No - 2.5 (1.3-81.6) 0.6 (0.1-0.8) - - 6 No - 5 (3.1-18.9) 0.9 (0.1-1.1) - - 7 No - 4.7 (2.7-11.7) 1.7 (0.9-2.7) - - 8 No - 460 (61.8-1244.8) 314.3 (4.6-1719.9) - - 634

635

636

637

638

639

640

641

642

643

644

645

646

647

648

649

650

651

652

653

654 655 Figures

656 657 Figure 1: Map of the River Tweed showing all the major tributaries as well as the migration route 658 downstream from the Yarrow Water. Grey boxes denote the release sites along with white circles 659 denoting the ALS positions and white diamonds for SEPA flow gauging stations (FGS). Black bars 660 indicate the sites of in-river structures.

661

662

663 664

665 Figure 2. Cumulative survival of acoustically tagged brown trout smolts migrating out to sea in 2010 666 and for three separate release groups in 2011. Black vertical bar represent weirs along the migration 667 route. * Measured from the furthest upstream release point down to the estuary. 668

669 Figure 3. Time spent by individual smolts at ALS positions (delay) that were within the 670 impoundment zones of in river structures (obstructed) compared with those that were 671 not (unobstructed). Data are presented as box plots, showing median, upper and lower 672 quartiles, upper and lower 5 percentiles, mild outliers (circles; Q3 + 1.5 × IQR) and 673 extreme outliers (asterisks; Q3 + 3 × IQR). In the 2010, panel medians are obscured by 674 other lines. Data do not include records from station 5 due to insufficient sample size. 675 676 677 Figure 4. Box plot displaying the median net ground speeds of tagged trout smolts moving through 678 each river section in both 2010 and 2011. Boxes represent upper and lower quartiles and T-bars 679 represent the upper and lower 5 percentiles and round dots signify outliers. *Section of river 680 between ALS stations, station 5 removed from analysis due to insufficient sample size. 70

2010 60

2011

) Q95 Flow 1

- 50 s 3 Q10 Flow

40

30

20 Mean daily flow daily (m Mean

10

0

14-Apr 18-Apr 22-Apr 26-Apr 30-Apr

01-Jun 05-Jun 09-Jun 13-Jun 17-Jun 21-Jun 25-Jun 29-Jun

04-May 08-May 12-May 16-May 20-May 24-May 28-May Date 681

682 Figure 5. Mean daily flows at the flow gauging station at Lindean on the Ettrick Water, reflecting 683 water flow at Murray's Cauld, during the period of study in both 2010 and 2011 as well as the Q95 684 and Q10 flows for the Lindean station.

50

45 Lower Yarrow Water

Lower Ettrick Water JUN)

- 40 Upper River Tweed 35

30

25

20

15

10

5 Days below Q95 flow flow (APR Days Q95 below

0

1986 1988 1990 1992 1994 1996 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1998 2000 2002 2004 2006 2008 2010

685 Year

686 Figure 6. Total number of days below Q95 flows for the smolt migration period 1 April to 30 May 687 between 1962 and 2011 on the lower Yarrow Water at the Philiphaugh flow gauging station, lower 688 Ettrick Water at the Lindean flow gauging station and the upper Tweed at the Boleside flow gauging 689 station.

690

6

)

1 R² = 0.5165

- 5

4 R² = 0.3348 3

2 2010 2011 1 Net ground speed (km (km h speedgroundNet Linear (2010) Linear (2011) 0 0 20 40 60 80 100 120 Mean flow (m3s-1) 691

692 Figure 7. The net ground speed (km h-1) of migrating smolts in relation to the estimated mean flow 693 conditions (m3s-1) during the period of transit throughout the migratory route. Flows are based upon 694 the nearest 15-minute gauged flow, at the closest gauging station.